Stable unCLIP checkpoints are finetuned from stable diffusion 2.1 checkpoints to condition on CLIP image embeddings. Stable unCLIP also still conditions on text embeddings. Given the two separate conditionings, stable unCLIP can be used for text guided image variation. When combined with an unCLIP prior, it can also be used for full text to image generation.
To know more about the unCLIP process, check out the following paper:
Hierarchical Text-Conditional Image Generation with CLIP Latents by Aditya Ramesh, Prafulla Dhariwal, Alex Nichol, Casey Chu, Mark Chen.
Stable unCLIP takes a noise_level
as input during inference. noise_level
determines how much noise is added
to the image embeddings. A higher noise_level
increases variation in the final un-noised images. By default,
we do not add any additional noise to the image embeddings i.e. noise_level = 0
.
Coming soon!
from diffusers import StableUnCLIPImg2ImgPipeline
from diffusers.utils import load_image
import torch
pipe = StableUnCLIPImg2ImgPipeline.from_pretrained(
"stabilityai/stable-diffusion-2-1-unclip", torch_dtype=torch.float16, variation="fp16"
)
pipe = pipe.to("cuda")
url = "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/stable_unclip/tarsila_do_amaral.png"
init_image = load_image(url)
images = pipe(init_image).images
images[0].save("variation_image.png")
Optionally, you can also pass a prompt to pipe
such as:
prompt = "A fantasy landscape, trending on artstation"
images = pipe(init_image, prompt=prompt).images
images[0].save("variation_image_two.png")
If you are short on GPU memory, you can enable smart CPU offloading so that models that are not needed immediately for a computation can be offloaded to CPU:
from diffusers import StableUnCLIPImg2ImgPipeline
from diffusers.utils import load_image
import torch
pipe = StableUnCLIPImg2ImgPipeline.from_pretrained(
"stabilityai/stable-diffusion-2-1-unclip", torch_dtype=torch.float16, variation="fp16"
)
# Offload to CPU.
pipe.enable_model_cpu_offload()
url = "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/stable_unclip/tarsila_do_amaral.png"
init_image = load_image(url)
images = pipe(init_image).images
images[0]
Further memory optimizations are possible by enabling VAE slicing on the pipeline:
from diffusers import StableUnCLIPImg2ImgPipeline
from diffusers.utils import load_image
import torch
pipe = StableUnCLIPImg2ImgPipeline.from_pretrained(
"stabilityai/stable-diffusion-2-1-unclip", torch_dtype=torch.float16, variation="fp16"
)
pipe.enable_model_cpu_offload()
pipe.enable_vae_slicing()
url = "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/stable_unclip/tarsila_do_amaral.png"
init_image = load_image(url)
images = pipe(init_image).images
images[0]
( prior_tokenizer: CLIPTokenizer prior_text_encoder: CLIPTextModelWithProjection prior: PriorTransformer prior_scheduler: KarrasDiffusionSchedulers image_normalizer: StableUnCLIPImageNormalizer image_noising_scheduler: KarrasDiffusionSchedulers tokenizer: CLIPTokenizer text_encoder: CLIPTextModelWithProjection unet: UNet2DConditionModel scheduler: KarrasDiffusionSchedulers vae: AutoencoderKL )
Parameters
CLIPTokenizer
) —
Tokenizer of class
CLIPTokenizer.
CLIPTextModelWithProjection
) —
Frozen text-encoder.
KarrasDiffusionSchedulers
) —
Scheduler used in the prior denoising process.
StableUnCLIPImageNormalizer
) —
Used to normalize the predicted image embeddings before the noise is applied and un-normalize the image
embeddings after the noise has been applied.
KarrasDiffusionSchedulers
) —
Noise schedule for adding noise to the predicted image embeddings. The amount of noise to add is determined
by noise_level
in StableUnCLIPPipeline.__call__
.
CLIPTokenizer
) —
Tokenizer of class
CLIPTokenizer.
CLIPTextModel
) —
Frozen text-encoder.
KarrasDiffusionSchedulers
) —
A scheduler to be used in combination with unet
to denoise the encoded image latents.
Pipeline for text-to-image generation using stable unCLIP.
This model inherits from DiffusionPipeline. Check the superclass documentation for the generic methods the library implements for all the pipelines (such as downloading or saving, running on a particular device, etc.)
(
prompt: typing.Union[str, typing.List[str], NoneType] = None
height: typing.Optional[int] = None
width: typing.Optional[int] = None
num_inference_steps: int = 20
guidance_scale: float = 10.0
negative_prompt: typing.Union[str, typing.List[str], NoneType] = None
num_images_per_prompt: typing.Optional[int] = 1
eta: float = 0.0
generator: typing.Optional[torch._C.Generator] = None
latents: typing.Optional[torch.FloatTensor] = None
prompt_embeds: typing.Optional[torch.FloatTensor] = None
negative_prompt_embeds: typing.Optional[torch.FloatTensor] = None
output_type: typing.Optional[str] = 'pil'
return_dict: bool = True
callback: typing.Union[typing.Callable[[int, int, torch.FloatTensor], NoneType], NoneType] = None
callback_steps: int = 1
cross_attention_kwargs: typing.Union[typing.Dict[str, typing.Any], NoneType] = None
noise_level: int = 0
prior_num_inference_steps: int = 25
prior_guidance_scale: float = 4.0
prior_latents: typing.Optional[torch.FloatTensor] = None
)
→
ImagePipelineOutput or tuple
Parameters
str
or List[str]
, optional) —
The prompt or prompts to guide the image generation. If not defined, one has to pass prompt_embeds
.
instead.
int
, optional, defaults to self.unet.config.sample_size * self.vae_scale_factor) —
The height in pixels of the generated image.
int
, optional, defaults to self.unet.config.sample_size * self.vae_scale_factor) —
The width in pixels of the generated image.
int
, optional, defaults to 20) —
The number of denoising steps. More denoising steps usually lead to a higher quality image at the
expense of slower inference.
float
, optional, defaults to 10.0) —
Guidance scale as defined in Classifier-Free Diffusion Guidance.
guidance_scale
is defined as w
of equation 2. of Imagen
Paper. Guidance scale is enabled by setting guidance_scale > 1
. Higher guidance scale encourages to generate images that are closely linked to the text prompt
,
usually at the expense of lower image quality.
str
or List[str]
, optional) —
The prompt or prompts not to guide the image generation. If not defined, one has to pass
negative_prompt_embeds
. instead. If not defined, one has to pass negative_prompt_embeds
. instead.
Ignored when not using guidance (i.e., ignored if guidance_scale
is less than 1
).
int
, optional, defaults to 1) —
The number of images to generate per prompt.
float
, optional, defaults to 0.0) —
Corresponds to parameter eta (η) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to
schedulers.DDIMScheduler, will be ignored for others.
torch.Generator
or List[torch.Generator]
, optional) —
One or a list of torch generator(s)
to make generation deterministic.
torch.FloatTensor
, optional) —
Pre-generated noisy latents, sampled from a Gaussian distribution, to be used as inputs for image
generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
tensor will ge generated by sampling using the supplied random generator
.
torch.FloatTensor
, optional) —
Pre-generated text embeddings. Can be used to easily tweak text inputs, e.g. prompt weighting. If not
provided, text embeddings will be generated from prompt
input argument.
torch.FloatTensor
, optional) —
Pre-generated negative text embeddings. Can be used to easily tweak text inputs, e.g. prompt
weighting. If not provided, negative_prompt_embeds will be generated from negative_prompt
input
argument.
str
, optional, defaults to "pil"
) —
The output format of the generate image. Choose between
PIL: PIL.Image.Image
or np.array
.
bool
, optional, defaults to True
) —
Whether or not to return a StableDiffusionPipelineOutput instead of a
plain tuple.
Callable
, optional) —
A function that will be called every callback_steps
steps during inference. The function will be
called with the following arguments: callback(step: int, timestep: int, latents: torch.FloatTensor)
.
int
, optional, defaults to 1) —
The frequency at which the callback
function will be called. If not specified, the callback will be
called at every step.
dict
, optional) —
A kwargs dictionary that if specified is passed along to the AttentionProcessor
as defined under
self.processor
in
diffusers.cross_attention.
int
, optional, defaults to 0
) —
The amount of noise to add to the image embeddings. A higher noise_level
increases the variance in
the final un-noised images. See StableUnCLIPPipeline.noise_image_embeddings
for details.
int
, optional, defaults to 25) —
The number of denoising steps in the prior denoising process. More denoising steps usually lead to a
higher quality image at the expense of slower inference.
float
, optional, defaults to 4.0) —
Guidance scale for the prior denoising process as defined in Classifier-Free Diffusion
Guidance. prior_guidance_scale
is defined as w
of equation 2. of
Imagen Paper. Guidance scale is enabled by setting
guidance_scale > 1
. Higher guidance scale encourages to generate images that are closely linked to
the text prompt
, usually at the expense of lower image quality.
torch.FloatTensor
, optional) —
Pre-generated noisy latents, sampled from a Gaussian distribution, to be used as inputs for image
embedding generation in the prior denoising process. Can be used to tweak the same generation with
different prompts. If not provided, a latents tensor will ge generated by sampling using the supplied
random generator
.
Returns
ImagePipelineOutput or tuple
~ pipeline_utils.ImagePipelineOutput
if return_dict
is
True, otherwise a tuple
. When returning a tuple, the first element is a list with the generated images.
Function invoked when calling the pipeline for generation.
Examples:
>>> import torch
>>> from diffusers import StableUnCLIPPipeline
>>> pipe = StableUnCLIPPipeline.from_pretrained(
... "fusing/stable-unclip-2-1-l", torch_dtype=torch.float16
... ) # TODO update model path
>>> pipe = pipe.to("cuda")
>>> prompt = "a photo of an astronaut riding a horse on mars"
>>> images = pipe(prompt).images
>>> images[0].save("astronaut_horse.png")
( slice_size: typing.Union[str, int, NoneType] = 'auto' )
Parameters
str
or int
, optional, defaults to "auto"
) —
When "auto"
, halves the input to the attention heads, so attention will be computed in two steps. If
"max"
, maximum amount of memory will be saved by running only one slice at a time. If a number is
provided, uses as many slices as attention_head_dim // slice_size
. In this case, attention_head_dim
must be a multiple of slice_size
.
Enable sliced attention computation.
When this option is enabled, the attention module will split the input tensor in slices, to compute attention in several steps. This is useful to save some memory in exchange for a small speed decrease.
Disable sliced attention computation. If enable_attention_slicing
was previously invoked, this method will go
back to computing attention in one step.
Enable sliced VAE decoding.
When this option is enabled, the VAE will split the input tensor in slices to compute decoding in several steps. This is useful to save some memory and allow larger batch sizes.
Disable sliced VAE decoding. If enable_vae_slicing
was previously invoked, this method will go back to
computing decoding in one step.
( attention_op: typing.Optional[typing.Callable] = None )
Parameters
Callable
, optional) —
Override the default None
operator for use as op
argument to the
memory_efficient_attention()
function of xFormers.
Enable memory efficient attention as implemented in xformers.
When this option is enabled, you should observe lower GPU memory usage and a potential speed up at inference time. Speed up at training time is not guaranteed.
Warning: When Memory Efficient Attention and Sliced attention are both enabled, the Memory Efficient Attention is used.
Examples:
>>> import torch
>>> from diffusers import DiffusionPipeline
>>> from xformers.ops import MemoryEfficientAttentionFlashAttentionOp
>>> pipe = DiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-2-1", torch_dtype=torch.float16)
>>> pipe = pipe.to("cuda")
>>> pipe.enable_xformers_memory_efficient_attention(attention_op=MemoryEfficientAttentionFlashAttentionOp)
>>> # Workaround for not accepting attention shape using VAE for Flash Attention
>>> pipe.vae.enable_xformers_memory_efficient_attention(attention_op=None)
Disable memory efficient attention as implemented in xformers.
Offloads all models to CPU using accelerate, reducing memory usage with a low impact on performance. Compared
to enable_sequential_cpu_offload
, this method moves one whole model at a time to the GPU when its forward
method is called, and the model remains in GPU until the next model runs. Memory savings are lower than with
enable_sequential_cpu_offload
, but performance is much better due to the iterative execution of the unet
.
Offloads all models to CPU using accelerate, significantly reducing memory usage. When called, the pipeline’s
models have their state dicts saved to CPU and then are moved to a torch.device('meta') and loaded to GPU only when their specific submodule has its
forward` method called.
( image_embeds: Tensor noise_level: int noise: typing.Optional[torch.FloatTensor] = None generator: typing.Optional[torch._C.Generator] = None )
Add noise to the image embeddings. The amount of noise is controlled by a noise_level
input. A higher
noise_level
increases the variance in the final un-noised images.
The noise is applied in two ways
In both cases, the amount of noise is controlled by the same noise_level
.
The embeddings are normalized before the noise is applied and un-normalized after the noise is applied.
( feature_extractor: CLIPImageProcessor image_encoder: CLIPVisionModelWithProjection image_normalizer: StableUnCLIPImageNormalizer image_noising_scheduler: KarrasDiffusionSchedulers tokenizer: CLIPTokenizer text_encoder: CLIPTextModel unet: UNet2DConditionModel scheduler: KarrasDiffusionSchedulers vae: AutoencoderKL )
Parameters
CLIPImageProcessor
) —
Feature extractor for image pre-processing before being encoded.
CLIPVisionModelWithProjection
) —
CLIP vision model for encoding images.
StableUnCLIPImageNormalizer
) —
Used to normalize the predicted image embeddings before the noise is applied and un-normalize the image
embeddings after the noise has been applied.
KarrasDiffusionSchedulers
) —
Noise schedule for adding noise to the predicted image embeddings. The amount of noise to add is determined
by noise_level
in StableUnCLIPPipeline.__call__
.
CLIPTokenizer
) —
Tokenizer of class
CLIPTokenizer.
CLIPTextModel
) —
Frozen text-encoder.
KarrasDiffusionSchedulers
) —
A scheduler to be used in combination with unet
to denoise the encoded image latents.
Pipeline for text-guided image to image generation using stable unCLIP.
This model inherits from DiffusionPipeline. Check the superclass documentation for the generic methods the library implements for all the pipelines (such as downloading or saving, running on a particular device, etc.)
(
image: typing.Union[torch.FloatTensor, PIL.Image.Image] = None
prompt: typing.Union[str, typing.List[str]] = None
height: typing.Optional[int] = None
width: typing.Optional[int] = None
num_inference_steps: int = 20
guidance_scale: float = 10
negative_prompt: typing.Union[str, typing.List[str], NoneType] = None
num_images_per_prompt: typing.Optional[int] = 1
eta: float = 0.0
generator: typing.Optional[torch._C.Generator] = None
latents: typing.Optional[torch.FloatTensor] = None
prompt_embeds: typing.Optional[torch.FloatTensor] = None
negative_prompt_embeds: typing.Optional[torch.FloatTensor] = None
output_type: typing.Optional[str] = 'pil'
return_dict: bool = True
callback: typing.Union[typing.Callable[[int, int, torch.FloatTensor], NoneType], NoneType] = None
callback_steps: int = 1
cross_attention_kwargs: typing.Union[typing.Dict[str, typing.Any], NoneType] = None
noise_level: int = 0
image_embeds: typing.Optional[torch.FloatTensor] = None
)
→
ImagePipelineOutput or tuple
Parameters
str
or List[str]
, optional) —
The prompt or prompts to guide the image generation. If not defined, either prompt_embeds
will be
used or prompt is initialized to ""
.
torch.FloatTensor
or PIL.Image.Image
) —
Image
, or tensor representing an image batch. The image will be encoded to its CLIP embedding which
the unet will be conditioned on. Note that the image is not encoded by the vae and then used as the
latents in the denoising process such as in the standard stable diffusion text guided image variation
process.
int
, optional, defaults to self.unet.config.sample_size * self.vae_scale_factor) —
The height in pixels of the generated image.
int
, optional, defaults to self.unet.config.sample_size * self.vae_scale_factor) —
The width in pixels of the generated image.
int
, optional, defaults to 20) —
The number of denoising steps. More denoising steps usually lead to a higher quality image at the
expense of slower inference.
float
, optional, defaults to 10.0) —
Guidance scale as defined in Classifier-Free Diffusion Guidance.
guidance_scale
is defined as w
of equation 2. of Imagen
Paper. Guidance scale is enabled by setting guidance_scale > 1
. Higher guidance scale encourages to generate images that are closely linked to the text prompt
,
usually at the expense of lower image quality.
str
or List[str]
, optional) —
The prompt or prompts not to guide the image generation. If not defined, one has to pass
negative_prompt_embeds
. instead. If not defined, one has to pass negative_prompt_embeds
. instead.
Ignored when not using guidance (i.e., ignored if guidance_scale
is less than 1
).
int
, optional, defaults to 1) —
The number of images to generate per prompt.
float
, optional, defaults to 0.0) —
Corresponds to parameter eta (η) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to
schedulers.DDIMScheduler, will be ignored for others.
torch.Generator
or List[torch.Generator]
, optional) —
One or a list of torch generator(s)
to make generation deterministic.
torch.FloatTensor
, optional) —
Pre-generated noisy latents, sampled from a Gaussian distribution, to be used as inputs for image
generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
tensor will ge generated by sampling using the supplied random generator
.
torch.FloatTensor
, optional) —
Pre-generated text embeddings. Can be used to easily tweak text inputs, e.g. prompt weighting. If not
provided, text embeddings will be generated from prompt
input argument.
torch.FloatTensor
, optional) —
Pre-generated negative text embeddings. Can be used to easily tweak text inputs, e.g. prompt
weighting. If not provided, negative_prompt_embeds will be generated from negative_prompt
input
argument.
str
, optional, defaults to "pil"
) —
The output format of the generate image. Choose between
PIL: PIL.Image.Image
or np.array
.
bool
, optional, defaults to True
) —
Whether or not to return a StableDiffusionPipelineOutput instead of a
plain tuple.
Callable
, optional) —
A function that will be called every callback_steps
steps during inference. The function will be
called with the following arguments: callback(step: int, timestep: int, latents: torch.FloatTensor)
.
int
, optional, defaults to 1) —
The frequency at which the callback
function will be called. If not specified, the callback will be
called at every step.
dict
, optional) —
A kwargs dictionary that if specified is passed along to the AttentionProcessor
as defined under
self.processor
in
diffusers.cross_attention.
int
, optional, defaults to 0
) —
The amount of noise to add to the image embeddings. A higher noise_level
increases the variance in
the final un-noised images. See StableUnCLIPPipeline.noise_image_embeddings
for details.
torch.FloatTensor
, optional) —
Pre-generated CLIP embeddings to condition the unet on. Note that these are not latents to be used in
the denoising process. If you want to provide pre-generated latents, pass them to __call__
as
latents
.
Returns
ImagePipelineOutput or tuple
~ pipeline_utils.ImagePipelineOutput
if return_dict
is
True, otherwise a tuple
. When returning a tuple, the first element is a list with the generated images.
Function invoked when calling the pipeline for generation.
Examples:
>>> import requests
>>> import torch
>>> from PIL import Image
>>> from io import BytesIO
>>> from diffusers import StableUnCLIPImg2ImgPipeline
>>> pipe = StableUnCLIPImg2ImgPipeline.from_pretrained(
... "fusing/stable-unclip-2-1-l-img2img", torch_dtype=torch.float16
... ) # TODO update model path
>>> pipe = pipe.to("cuda")
>>> url = "https://raw.githubusercontent.com/CompVis/stable-diffusion/main/assets/stable-samples/img2img/sketch-mountains-input.jpg"
>>> response = requests.get(url)
>>> init_image = Image.open(BytesIO(response.content)).convert("RGB")
>>> init_image = init_image.resize((768, 512))
>>> prompt = "A fantasy landscape, trending on artstation"
>>> images = pipe(prompt, init_image).images
>>> images[0].save("fantasy_landscape.png")
( slice_size: typing.Union[str, int, NoneType] = 'auto' )
Parameters
str
or int
, optional, defaults to "auto"
) —
When "auto"
, halves the input to the attention heads, so attention will be computed in two steps. If
"max"
, maximum amount of memory will be saved by running only one slice at a time. If a number is
provided, uses as many slices as attention_head_dim // slice_size
. In this case, attention_head_dim
must be a multiple of slice_size
.
Enable sliced attention computation.
When this option is enabled, the attention module will split the input tensor in slices, to compute attention in several steps. This is useful to save some memory in exchange for a small speed decrease.
Disable sliced attention computation. If enable_attention_slicing
was previously invoked, this method will go
back to computing attention in one step.
Enable sliced VAE decoding.
When this option is enabled, the VAE will split the input tensor in slices to compute decoding in several steps. This is useful to save some memory and allow larger batch sizes.
Disable sliced VAE decoding. If enable_vae_slicing
was previously invoked, this method will go back to
computing decoding in one step.
( attention_op: typing.Optional[typing.Callable] = None )
Parameters
Callable
, optional) —
Override the default None
operator for use as op
argument to the
memory_efficient_attention()
function of xFormers.
Enable memory efficient attention as implemented in xformers.
When this option is enabled, you should observe lower GPU memory usage and a potential speed up at inference time. Speed up at training time is not guaranteed.
Warning: When Memory Efficient Attention and Sliced attention are both enabled, the Memory Efficient Attention is used.
Examples:
>>> import torch
>>> from diffusers import DiffusionPipeline
>>> from xformers.ops import MemoryEfficientAttentionFlashAttentionOp
>>> pipe = DiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-2-1", torch_dtype=torch.float16)
>>> pipe = pipe.to("cuda")
>>> pipe.enable_xformers_memory_efficient_attention(attention_op=MemoryEfficientAttentionFlashAttentionOp)
>>> # Workaround for not accepting attention shape using VAE for Flash Attention
>>> pipe.vae.enable_xformers_memory_efficient_attention(attention_op=None)
Disable memory efficient attention as implemented in xformers.
Offloads all models to CPU using accelerate, reducing memory usage with a low impact on performance. Compared
to enable_sequential_cpu_offload
, this method moves one whole model at a time to the GPU when its forward
method is called, and the model remains in GPU until the next model runs. Memory savings are lower than with
enable_sequential_cpu_offload
, but performance is much better due to the iterative execution of the unet
.
Offloads all models to CPU using accelerate, significantly reducing memory usage. When called, the pipeline’s
models have their state dicts saved to CPU and then are moved to a torch.device('meta') and loaded to GPU only when their specific submodule has its
forward` method called.
( image_embeds: Tensor noise_level: int noise: typing.Optional[torch.FloatTensor] = None generator: typing.Optional[torch._C.Generator] = None )
Add noise to the image embeddings. The amount of noise is controlled by a noise_level
input. A higher
noise_level
increases the variance in the final un-noised images.
The noise is applied in two ways
In both cases, the amount of noise is controlled by the same noise_level
.
The embeddings are normalized before the noise is applied and un-normalized after the noise is applied.